-
相较于传统面阵凝视及微振镜扫描型的光子计数激光雷达只需提取每一像素点的累计峰值即可获得固定目标表面的三维结构信息,此激光雷达系统采用多元InGaAs/InP SPAD配合机械扫描的方式实时动态获取大角度大尺度范围目标表面的三维信息,无法通过提取光子计数峰值的方式获取目标信号,需要对获取的光子点云图像进行滤波去噪。为了提高信噪比,需要尽可能提高目标的扫描次数即点云密度,而目标越远,点云密度越小,相邻两次探测的空间间隔由公式(1)表示:
$$ \Delta L = \frac{{\omega R}}{{{f_L}}} $$ (1) 式中:
$\omega $ 为电机转速;$R$ 为目标距离;${f_L}$ 为激光重频。采用自由运转模式SPAD可以大大提高探测器的探测频率。自由运转模式是单光子探测器针对目标距离未知情况的一种探测方式,SPAD始终处于雪崩就绪状态,等待光子信号触发,当探测到光子事件发生雪崩效应之后,通过主被动混合淬灭模式进行雪崩淬灭,以实现单光子探测器的高效率连续探测。但由于受到噪声和死时间的影响,单光子探测器无法保证每次探测都能产生有效的回波信号。光子计数激光雷达的噪声主要来自单光子探测器的背景光噪声和探测器本身的暗计数两部分,两者是完全独立的两种随机过程,所以可以用下式表示[15]:
$$ {N_t} = {N_b} + {N_d} $$ (2) 式中:
$ {N_t} $ 为总的平均噪声光电子数,代表了在单位时间内探测器响应的噪声光电子数的平均值;$ {N_b} $ 为背景光噪声的平均光电子数;$ {N_d} $ 为探测器的暗计数。暗计数$ {N_d} $ 是由于热运动、隧道效应和势阱捕获效应等原因导致在没有任何光照时在电场作用下也会产生自触发的雪崩效应,其中由于势阱捕获载流子后释放产生的后脉冲效应是暗计数的主要原因之一,也是限制InGaAs/InP SPAD性能的主要因素[16]。为了抑制InGaAs SPAD后脉冲效应,需要通过延长死时间来使被捕获的载流子从倍增区域释放,从而减小触发后脉冲的可能性。然而,施加较长的死时间限制了探测器的饱和计数率从而降低了探测效率,所以死时间大小的选择是后脉冲抑制和最大计数率之间的权衡[17]。相比于Si基器件,III-V族器件中具有非常高密度的少数载流子俘获能级,这也是InGaAs/InP器件总体性能远低于Si基器件的主要原因之一[18]。背景光噪声$ {N_b} $ 包括背景自身的热辐射光噪声和太阳背景辐射光噪声,而由于文中系统工作在近红外波段,背景自身的热辐射光噪声可以忽略不计,而太阳背景辐射光噪声包括目标对太阳光的反射和太阳光的大气散射两部分,在系统的探测距离范围内,大气透射度较高,大气后向散射噪声趋近于零,因此可忽略大气的后向散射产生的日光噪声,采用下式表示系统背景噪声模型[19]:$$ {N_b} = \rho {T_a}{T'_a}{\eta _r}\theta _r^2{A_r}{H_\lambda }\Delta \lambda \cos \theta \frac{{{\eta _q}}}{{hv}} $$ (3) 式中:
$ {N_b} $ 为单位时间内探测器探测到的背景噪声平均光电子数;$ {\eta _q} $ 为探测器的量子效率;$\; \rho $ 为激光波长下第目标反射率;$ {T_a} $ 为目标到激光雷达系统的大气透过率;$ {T'_a} $ 为太阳到目标的大气透过率;$ {\eta _r} $ 为激光雷达的光学接收效率;$ \theta _r^{} $ 为目标的视场角;$ {A_r} $ 为接收光学系统的孔径;$ {H_\lambda } $ 为太阳光对地面的光谱辐照度;$ \Delta \lambda $ 为滤波带宽;$ \theta $ 为阳光照射方向和目标表面的法向夹角。图1为自由运转模式单光子探测电路的原理结构,主要包括单光子探测器、温控电路、偏压电路、淬灭恢复电路、鉴别整形电路。为了避免淬灭过程中恢复时间长的问题,采用主动淬灭型电路方案,即采用比较器等外部器件构建一个对SPAD偏置电压的负反馈机制。
SPAD的工作性能(探测效率和暗计数)受工作条件(制冷温度和偏置电压)的影响,需要对探测电路的最佳工作条件进行选择以达到最佳工作性能。单光子探测器在实际应用中应综合考虑探测效率和暗计数指标。
-
当前针对光子计数激光雷达的数据特点,后续数据处理的难点主要集中在光子点云去噪滤波算法方面。大量的背景噪声、暗计数等噪声点随机分在整个测量范围内,信号点则以线状连续集聚的形式分布在特定的距离上,且密度明显大于噪声点的密度。目前对光子点云的去噪算法可以分为两类,一类是基于直方图、概率统计、聚类等的统计分析方法[20–23],另一类是基于边缘检测、图像分割等的图像处理技术[23–25]。每种算法各有优势,所适用的场景和解决的问题不尽相同。由于文中系统的噪声密度较大、分布不均且后脉冲噪声明显,单一算法无法获得最佳的处理效果。文中采用分步由粗到精的噪声滤除策略,结合多种算法的优势,针对地面远距离孤立目标的应用场景,提出一种更为鲁棒的噪声滤除方法。
针对地面远距离孤立目标的应用场景,采用分步由粗到精的噪声滤除策略旨在从以下三个方面进行算法设计。第一,目标可能出现在测量范围的任意距离处,没有任何关于目标距离的先验信息。同时,噪声充满了整个测量范围,因此首先考虑剔除大部分的噪声点以降低后续数据处理量;第二,在地面测量平台下,不同于机载或星载平台,目标并不是以连续线状形式呈现的,而是孤立地存在于自身的空间范围内,因此,精细化处理可根据目标的孤立聚集形态开展;第三,系统存在明显的后脉冲现象,表现为在目标后方存在和目标形态类似的伪信号,需要根据其特征进行鉴别并剔除。
基于以上分析,设计了适合该场景下的点云滤波及后脉冲预处理算法,图2为算法的处理流程图。具体步骤及结果如下:
(1)首先,读取原始光子点云数据,按照时间距离组织成二维平面数据,横轴为测量时间,纵轴为测量距离;
(2)选取合适的时间累积间隔Δt,将数据进行垂直切片,在每一个切片内进行距离向的直方图统计,将大于均值2倍标准差的距离保留,作为初步的目标点候选集合;
(3)得到粗去噪结果后,进一步使用DBSCAN(Density-Based Spatial Clustering of Applications with Noise)进行基于密度的空间聚类,聚类后可将其中具有高密度相连的区域划分为同一集合簇,并剔除粗去噪结果中的大部分点数较少的点。
然而基于密度的聚类结果并不能将后脉冲信号识别出,后脉冲噪点簇将被误认为是低密度的目标簇。为了进一步鉴别真实目标信号之后的后脉冲簇,对每一个聚类的点簇集合沿法线向后移动一段距离寻找后脉冲簇,并判断是否存在后脉冲簇,存在时则将后脉冲簇删除。
-
激光雷达系统采用光子计数体制,主要技术包括窄脉冲高重频短波红外激光技术、InGaAs/InP SPAD单光子探测及主动淬灭技术、光子计数技术及高精度时间测量技术等,组成部分主要包括:激光器模块、收发光学模块、单光子探测器模块、高精度时间测量模块、扫描电机模块、电源模块。系统的主要参数及结构如表1和图3所示。
表 1 激光雷达系统主要参数
Table 1. Key parameters of the lidar system
Devices Parameter Value Laser Wavelength 1550 nm Total energy 100 μJ Energy per beam 0.5 μJ Pulse width 1 ns Pulse frequency 20 kHz Beam splitting scale 128×1 Emission optics Divergence angle 0.15 mrad Maximum measured distance >3 km Included angle between beams 1.37 mrad Receiving optics FOV 10° IFOV 0.3 mrad Aperture 34 mm F# 4.9 SPAD Dark count 20 kHz Dead time 200 ns Detection efficiency 3%@1550 nm 1550 nm光纤激光器产生的激光经过扩束准直后通过衍射光学元件 DOE进行分束,光纤激光器的输出光纤芯径 25 μm,数值孔径 0.10,系统发射波束为 128,每一波束的发散角为 0.15 mrad,相邻波束之间的夹角为 13.7 mrad,衍射元件的尺寸为 38 mm,有效工作范围为 34 mm。接收望远镜采用像方远心的设计,将 128 束回波与线阵光纤进行耦合,之后每根光纤经过中心波长 1550.12 nm、带宽为 0.2 nm 的窄带滤光片滤除背景噪声,最后通过多模光纤与探测器进行耦合。
激光雷达系统采用的单光子探测器为中国科学院上海技术物理研究所自研的1550 nm波段InGaAs/InP盖革模式雪崩光电二极管,其在1550 nm波段探测效率>3%,内置有三级TEC制冷器,采用 TO-66的 封装结构,耦合50/100 μm多模尾纤,内置有效探测面直径35 μm的盖革模式雪崩光电二极管芯片。接收光纤经光纤滤波器后,各根光纤与单元探测器逐一对准。探测电路集成了主动淬灭、TEC温控、APD偏压和信号读出等功能。探测器工作在自由运转模式下,在每一只InGaAs/InP单光子探测器上独立进行主动淬灭控制,更快地进行雪崩抑制过程,进一步降低噪声对探测概率的影响,同时降低死时间并提高饱和计数率。TEC温控模块将探测器器件的温度稳定在−40~−20 ℃的可调区间,偏压控制单元可以提供稳定可调的独立偏压调节。
通过MODTRAN软件分析,1550 nm波段大气透过率约为80%,在日光条件下的背景光谱辐照度为H1550 nm=0.26 W/(m2·nm);在夜晚时,背景光谱辐照度为H1550 nm=1.5×10−6 W/(m2·nm)。将以上参数代入公式(3)计算可知,对于30%反射率的目标表面,系统的日光条件背景光子噪声强度约为3 MHz,相当于每个测距周期约150个光子事件;夜晚时背景光子噪声强度在1 kHz以下,相当于每个测距周期约0.05个光子事件。
Research on linear array scanning lidar and photon signal processing technology based on InGaAs single-photon detector
-
摘要: 随着探测体系的发展,基于单光子探测技术的光子计数激光雷达受到了广泛关注,有效降低了系统对激光功率的需求,广泛应用在远距离测距及成像领域。针对激光雷达在人眼安全波段的工作需求,基于自由运转模式InGaAs/InP SPAD单光子探测器设计了一套多元收发的远程线阵光子计数激光雷达扫描成像原型系统,对探测器在日光背景下的探测概率影响因素展开了分析,配合主动淬灭电路设计及工作温度、偏压调整获得了系统的最佳工作点,并针对扫描视场中孤立目标特征采用了点云滤波及后脉冲预处理算法,将单个接收通道的原始数据率由200 kbps量级降低至小于1 kbps。与记录单次回波相比,单个测距周期记录四次回波可将有效数据量提升约5%。同时也对探测器的噪声及后脉冲等特性进行了分析。该系统工作波段为1 550 nm,探测器线阵规模可达到128元,激光重频为20 kHz,可在2 s内实现水平200°范围内的激光三维成像,作用距离>3 km。经过成像算法处理,该系统在日光条件下成功实现多距离目标三维成像,成像目标清晰。
-
关键词:
- 激光雷达 /
- 光子计数 /
- 三维成像 /
- 单光子探测 /
- InGaAs/InP SPAD
Abstract:Objective With the development of the detection system, the photon-counting imaging lidar based on single-photon detection technology has greatly improved the detection sensitivity of the echo optical signal, effectively reduced the demand of the system for laser power, and made it possible for miniaturized, long-distance, high-resolution, and high-precision laser 3D imaging equipment, which is widely used in the field of long-distance ranging and imaging and has become a research hotspot.Since the number of echo photons in long-distance laser detection is only the order of single photons, the detection performance of the detector is highly required. So at present, most photon-counting lidar remote imaging generally adopts the method of area array staring detection or unit micro-mirror scanning to accumulate the fixed target for a long time to improve the signal-to-noise ratio, which is not conducive to real-time dynamic measurement of large-scale and large-range targets.Although there are many schemes using SPAD (Single-photon Avalanche Diode) to carry out ranging and imaging experiments at home and abroad, Si-based SPAD in the visible band is mainly used. Compared with InGaAs/InP SPAD, the dark count, detection efficiency, afterpulse probability, dead time and other indicators are not ideal. In order to achieve higher single-photon detection performance in near-infrared band, InGaAs/InP SPAD mainly adopts gated quenching mode, which is more suitable for the situation where the target distance is known, while not suitable for the situation where the target range is large, and the relevant research on the range measurement and imaging experiment using the active quenching mode InGaAs/InP SPAD of the free-running system is less, which is only at the stage of principle prototype. Therefore, this study proposes a linear array imaging lidar scheme based on InGaAs/InP single-photon detector. Methods Aiming at the working requirements of the eye safety band, based on the free-running mode InGaAs/InP SPAD, a set of remote linear array photon-counting lidar scanning imaging prototype system with multiple transceivers is designed, 128 units of InGaAs/InP SPAD are spliced into a linear array arrangement, the working band of the system is 1 550 nm, the laser repetition frequency is 20 kHz and the laser scanning imaging in the horizontal 200° range is realized through scanning in 2 seconds (Fig.3, Tab.5).At the same time, the factors affecting the detection probability of the detector in the sunlight background are analyzed, and the optimal working point of the system is obtained by combining with the active quenching circuit design (Fig.1) and the adjustment of the working temperature and bias voltage. Point cloud filtering and afterpulse preprocessing algorithms are used to reduce the original data rate of a single receiving channel and characterize isolated targets in the scanning field of view (Fig.2). Results and Discussions By analyzing the characteristics of afterpulse and noise, it can be seen that the background noise signal, including the background noise and the dark count signal, is randomly distributed in the whole space. With the increase of the number of echo statistics per unit ranging period, the background noise will increase. Behind the light count signal, there are two levels of obvious afterpulse signal, and the afterpulse caused by the dark count will also lead to the increase of the background noise (Fig.5). After the processing of the point cloud filtering and afterpulse preprocessing algorithm, the original data rate of a single receiving channel is reduced from 200 kbps to less than 1 kbps, and the obvious afterpulse signal behind the target point cloud is removed. Compared with recording single echo, recording four echoes in a single ranging cycle can increase the effective data volume by about 5%. After the processing of the imaging algorithm, the system successfully realizes three-dimensional imaging of multi-range targets under daylight conditions, the maximum detection distance is more than 3 km, and the imaging targets are clear (Fig.7). Conclusions In this study, a long-distance imaging lidar system based on photon-counting detection technology is designed, and the detection performance of single-photon detector is studied. After that, the noise and after-pulse characteristics, imaging clarity, and other indicators of the lidar are verified by fixed-point ranging experiments and scanning imaging experiments. The experimental results show that the system can successfully detect multi-range target information, and can detect long-distance targets of more than 3 km. The final reconstructed image is clear, and the noise suppression effect is perfect. The system is suitable for three-dimensional point cloud imaging of long-distance targets. However, due to the significant after-pulse effect of InGaAs/InP SPAD, the increase of dark count will seriously affect the detection performance. The afterpulse and background noise can be significantly filtered by point cloud filtering and afterpulse preprocessing algorithm, and the point cloud data can be compressed for subsequent processing. -
Key words:
- lidar /
- photon counting /
- 3D imaging /
- single-photon detection /
- InGaAs/InP SPAD
-
表 1 激光雷达系统主要参数
Table 1. Key parameters of the lidar system
Devices Parameter Value Laser Wavelength 1550 nm Total energy 100 μJ Energy per beam 0.5 μJ Pulse width 1 ns Pulse frequency 20 kHz Beam splitting scale 128×1 Emission optics Divergence angle 0.15 mrad Maximum measured distance >3 km Included angle between beams 1.37 mrad Receiving optics FOV 10° IFOV 0.3 mrad Aperture 34 mm F# 4.9 SPAD Dark count 20 kHz Dead time 200 ns Detection efficiency 3%@1550 nm -
[1] Dabney P, Harding D, Abshire J, et al. The slope imaging multi-polarization photon-counting lidar: Development and performance results[C]//2010 IEEE International Geoscience and Remote Sensing Symposium. New York: IEEE, 2010: 653-656. [2] Jingjing G, Xiaoyan F, Peng G, et al. High-resolution three-dimensional imaging based on all-fiber photon-counting Lidar system [J]. Infrared and Laser Engineering, 2021, 50(7): 20210162. (in Chinese) doi: 10.3788/IRLA20210162 [3] Marino R M, Davis W R, Rich G C, et al. High-resolution 3D imaging laser radar flight test experiments[C]//Laser Radar Technology and Applications X. SPIE, 2005, 5791: 138-151. [4] Marino R M, Davis W R. Jigsaw: a foliage-penetrating 3D imaging laser radar system [J]. Lincoln Laboratory Journal, 2005, 15(1): 23-36. [5] Knowlton R. Airborne ladar imaging research testbed[R]. Lexington, MA, United States: MIT Lincoln Laboratory, 2011. [6] Gray G. High Altitude Lidar Operations Experiment (HALOE)–Part 1, system design and operation[C]//Proc of Military Sensing Symposium, Active Electro-Optic Systems, 2011: 12-14. [7] Albota M A, Gurjar R S, Mangognia A V D, et al. The airborne optical systems testbed (AOSTB)[R]. Lexington, MA, United States: MIT Lincoln Laboratory, 2017. [8] Clifton W E, Steele B, Nelson G, et al. Medium altitude airborne Geiger-mode mapping LIDAR system[C]//Laser Radar Tech-nology and Applications XX and Atmospheric Propagation XII. SPIE, 2015, 9465: 39-46. [9] Sun Jianfeng, Jiang Peng, Zhang Xiuchuan, et al. Experimental research of 32×32 InGaAs Gm-APD arrays laser active imaging [J]. Infrared and Laser Engineering, 2016, 45(12): 1206006. (in Chinese) doi: 10.3788/IRLA201645.1206006 [10] Du B, Pang C, Wu D, et al. High-speed photon-counting laser ranging for broad range of distances [J]. Scientific Reports, 2018, 8(1): 1-6. doi: 10.1038/s41598-018-22675-1 [11] Li Z P, Ye J T, Huang X, et al. Single-photon imaging over 200 km [J]. Optica, 2021, 8(3): 344-349. doi: 10.1364/OPTICA.408657 [12] Itzler M A, Entwistle M, Owens M, et al. Geiger-mode avalanche photodiode focal plane arrays for three-dimensional imaging LADAR[C]//Infrared Remote Sensing and Instrumentation XVIII. SPIE, 2010, 7808: 75-88. [13] Jiang X, Itzler M, O’Donnell K, et al. InP-based single-photon detectors and Geiger-mode APD arrays for quantum communications applications [J]. IEEE Journal of Selected Topics in Quantum Electronics, 2014, 21(3): 5-16. doi: 10.1109/JSTQE.2014.2358685 [14] Itzler M A, Jiang X, Entwistle M, et al. Single-photon detectors based on InP avalanche diodes: status and prospects [J]. Advanced Photon Counting Techniques II, 2010, 7681: 195-206. [15] Libing H. Research on key technologies of photon counting imaging lidar in moving conditions[D]. Beijing: University of Chinese Academy of Sciences, 2013. (in Chinese) [16] Cova S, Ghioni M, Lotito A, et al. Evolution and prospects for single-photon avalanche diodes and quenching circuits [J]. Journal of Modern Optics, 2004, 51(9-10): 1267-1288. doi: 10.1080/09500340410001670839 [17] Buller G S, Collins R J. Single-photon detectors for infrared wavelengths in the range 1-1.7 μm[M]//Advanced Photon Counting. Cham: Springer, 2014: 43-69. [18] Lacaita A, Zappa F, Cova S, et al. Single-photon detection beyond 1 μm: performance of commercially available InGaAs/InP detectors [J]. Applied Optics, 1996, 35(16): 2986-2996. doi: 10.1364/ao.35.002986 [19] Milstein A B, Jiang L A, Luu J X, et al. Acquisition algorithm for direct-detection ladars with Geiger-mode avalanche photodiodes [J]. Applied Optics, 2008, 47(2): 296-311. doi: 10.1364/AO.47.000296 [20] Brunt K M, Neumann T A, Walsh K M, et al. Determination of local slope on the greenland ice sheet using a multibeam photon-counting lidar in preparation for the ICESat-2 mission [J]. IEEE Geoscience and Remote Sensing Letters, 2013, 11(5): 935-939. [21] Horan K H, Kerekes J P. An automated statistical analysis approach to noise reduction for photon-counting lidar systems[C]//IGARSS, 2013: 4336-4339. [22] Zhang J, Kerekes J, Csatho B, et al. A clustering approach for detection of ground in micropulse photon-counting LiDAR altimeter data[C]//2014 IEEE Geoscience and Remote Sensing Symposium. New York: IEEE, 2014: 177-180. [23] Magruder L A, Wharton III M E, Stout K D, et al. Noise filtering techniques for photon-counting ladar data[C]//Laser Radar Technology and Applications XVII. SPIE, 2012, 8379: 237-245. [24] Awadallah M, Ghannam S, Abbott L, et al. Active contour models for extracting ground and forest canopy curves from discrete laser altimeter data[C]//Proceedings: 13th International Conference on LiDAR Applications for Assessing Forest Ecosystems, 2013: 129-136. [25] Awadallah M S T. Image analysis techniques for LiDAR point cloud segmentation and surface estimation[D]. Virginia, US: Virginia Tech, 2016.